def du(summarize, human_readable, s3_paths):
"""
display disk usage statistics
"""
size_formatter = magic.human_bytes if human_readable else lambda s: s
for s3_path in s3_paths:
totals = {}
# In recursive tallying of totals this key is the stop condition.
stop_path = '/' + s3_util.bucket_and_key(s3_path)[1].rstrip('/')
def tally(path_segment, size):
if not summarize or path_segment == stop_path:
# If the summarize option is present, we only care about the grand total.
if path_segment not in totals:
totals[path_segment] = 0
totals[path_segment] += size
if path_segment != stop_path:
parent_dir = path_segment.rsplit('/', 1)[0]
if parent_dir == '':
# Edge case when totalling the entire bucket.
parent_dir = '/'
tally(parent_dir, size)
for obj in s3_util.keys(s3_path):
# usage for all given paths, and recursively for directories (excludes individual files)
dbg('adding {}', obj)
dir_key = '/' + obj['key'].rsplit('/', 1)[0]
tally(dir_key, obj['len'])
for path, total in sorted(totals.items()):
out('{}\t{}', size_formatter(total), path)
def scanb(start="", end=None):
bucket_start, first_bucket_key_prefix = bucket_and_key(start)
dbg("start b:{} k:{}", bucket_start, first_bucket_key_prefix if first_bucket_key_prefix else "(unbounded)")
bucket_end = None
last_bucket_key_end = None
if end:
bucket_end, last_bucket_key_end = bucket_and_key(end)
if not start <= end:
err("start must be lexically before end")
dbg("end b:{} k:{}", bucket_end, last_bucket_key_end if last_bucket_key_end else "(unbounded)")
s3 = boto3.resource("s3")
def bucket_gen():
_buckets = s3.buckets.all()
for _bucket in _buckets:
if _bucket.name < bucket_start:
continue
if bucket_end is not None and _bucket.name >= bucket_end:
break
yield _bucket
def returned_generator():
buckets = list(bucket_gen())
for index, bucket in enumerate(buckets):
key_start = first_bucket_key_prefix if index == 0 else ""
key_end = last_bucket_key_end if index == len(buckets) - 1 else None
yield scank(bucket.name, key_start, key_end)
return magic.flatten(returned_generator())
def returned_gen():
list_objects = s3.get_paginator("list_objects")
response_iterator = list_objects.paginate(Bucket=bucket_name, Prefix=key_prefix, Delimiter="/")
try:
for entry in magic.flatten(map(generate_objs, response_iterator)):
yield entry
except SignalUnslashedDirectory:
dbg("Saw unslashed directory at {}. Restarting paginator.", key_prefix)
response_iterator = list_objects.paginate(
Bucket=bucket_name,
Prefix=key_prefix + "/",
# Step into that dir! The user left off the trailing '/'.
Delimiter="/",
)
for entry in magic.flatten(map(generate_objs, response_iterator)):
yield entry
def scank(bucket_name, key_start="", key_end=None):
dbg("scanning b:{} s:{} e:{}", bucket_name, key_start, str(key_end))
prefix = ""
if key_start and key_end:
# Skip some unnecessary scanning by only going over the common prefix
prefix = os.path.commonprefix([key_start, key_end])
s3 = boto3.resource("s3")
bucket = s3.Bucket(bucket_name)
def key_gen():
_objects = bucket.objects.filter(Prefix=prefix)
for _obj in _objects:
if _obj.key < key_start:
continue
if key_end is not None and _obj.key >= key_end:
break
yield _obj
return ({"bucket": obj.bucket_name, "key": obj.key} for obj in key_gen())
